🔍 Read the full analysis: Jev Can See Now. I Pointed It At 30 Of My Own Images. on ThorstenMeyerAI.com
Get audio and creator gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
TypeSafe released Jev 1.14 on October 14, adding opt-in image inputs and a bounding-box response type. Thorsten Meyer tested the capabilities on 30 images from his own site; product-image matching was strong, but the supplied account cuts off before reporting the full news-image results. TypeSafe says the new features are not ready for production use without further evaluation.
TypeSafe made image support available through a request header, TypeSafe-Danger-Zone-Images: true. Images can be supplied as part of the model’s state, question instructions or criteria for choices and scoring. The company says this works with both Jev 1.13.0 and 1.14.0. A separate box response type, which returns one bounding box and a confidence score for a specified target, is limited to Jev 1.14.0.
TypeSafe has kept the new capabilities off by default. In its release post, the company said there was “still way too much jaggedness for us to vouch for these for production use cases.” It described the opt-in mechanism as a way to let builders retain existing behavior while allowing others to test the features. Image input accepts PNG and JPEG stills, with stated dimensions of 256 to 1,568 pixels per side, a 16 MiB limit per file and a 32 MiB limit per request. The release notes say a request fails if it contains an invalid image.
Meyer tested the model on 30 recent featured images from ThorstenMeyerAI.com: 10 product roundups, 15 news posts and five essays. He asked whether each image fit its own headline, a randomly selected other headline and an unrelated plant-care headline. The 10 product images were also used in a question asking Jev to identify the product. Meyer says the test involved about 60 requests across Jev 1.13.0 and 1.14.0. His report gives product-image results but the supplied text ends before detailing the news-image findings.
Jev 1.14 · images + box
Jev can see now. Here is what 30 of my images showed.
I asked about every featured image three times: does it fit its own headline, a random other post’s headline, an off-topic headline? Both Jev versions, about 60 requests.
Does this image fit the headline?
Median “yes” probability on Jev 1.14, by kind of post
- Product roundups10 images0.760.061.13: 0.79 / 0.05
- News posts15 images0.330.161.13: 0.51 / 0.25
- My essays5 images0.510.161.13: 0.62 / 0.33
Off-topic headline (“How to Repot a Monstera Plant”): never above 0.10 on either version, 60 of 60 rejected.
Why news is weak: almost all news images show the same scene, people at desks looking at monitors. Jev could not tell them apart, and it was right not to.
1.14 is stricter than 1.13
Same 30 images, same questions
| Result | 1.14 | 1.13 |
|---|---|---|
| Fits own headline (≥ 0.5) | 15/30 | 23/30 |
| Fits a wrong headline (≥ 0.5) | 1/30 | 4/30 |
| Own beats random other (AUC) | 0.91 | 0.89 |
| Names the product (10 options) | 9/10 | 10/10 |
1.14’s one product miss: mesh system read as “Wi-Fi 7 router”, confidence 0.60. Re-tune thresholds before jev-latest moves.
What shipped on Oct 14
- headerImagesin state, instructions or criteria · 1.13 + 1.14 ·
TypeSafe-Danger-Zone-Images - 1.14 onlyBox typeone bounding box + confidence. 6 of 6 clear targets boxed (0.66–0.90); an absent bicycle still got a box (0.03)
- headerLong context64k / 32k → 262k tokens, slower and less accurate
- opt-inHelp find weaknessesqueries used for testing, not training
now: jev-preview = 1.14~Oct 21: jev-latest → 1.14notice: 60 days (1.13), 30 days (1.14)
Source: own test on October 15, 2026. 30 featured images from ThorstenMeyerAI.com (10 product roundups, 15 news posts, 5 essays), JPEG at up to 1,568 px, jev-1.14.0 and jev-1.13.0 with the image header. Bars show medians. AUC = probability that an image’s own headline scores higher than a random other headline. Features, prices and limits: TypeSafe blog post “Jev 1.14” and docs.typesafe.ai, as of October 15, 2026.
Powered by Thorsten Meyer AI
Image Checks Move Beyond Text
The update gives developers a way to apply Jev’s structured decision format to visual inputs, including checks that might be difficult to standardize manually. A publisher, for example, could ask whether an image matches a headline, while an application could request a box around a named object. Jev returns constrained responses, such as a yes-or-no probability or a confidence-scored selection, rather than composing an open-ended description.
Meyer’s product-image test offers an early, limited example of where that approach may work. He reports that Jev 1.14 matched all 10 product images strongly to their own headlines compared with the alternative headlines. It also identified the correct product in nine of 10 images. Those results come from a small sample on one website, however; they do not establish accuracy across other image types, datasets or real-world workflows. TypeSafe’s production warning makes independent testing especially relevant before relying on the outputs.
As an affiliate, we earn on qualifying purchases.
Jev’s Text-First Design
Jev is a decision model that responds to typed questions with structured outputs, including probabilities for yes-or-no questions, selections from a list and scores on a scale. Meyer’s earlier coverage described it as a model for classification and other bounded decisions rather than sentence writing. Before this release, the questions in that workflow had to concern text.
TypeSafe says image support also works with jev-1.13.0, describing that model as multimodal even before the new release. The 1.14 release also includes broader model improvements, with multilingual performance expected by TypeSafe to benefit most. The new version is available as jev-preview; the company said jev-latest would move to it about a week after release, around October 21.
The release includes other changes separate from image processing. A long-context header raises the stated request limit from 64,000 to 262,000 tokens and the per-question limit from 32,000 to 262,000. TypeSafe still advises against using the larger limits because reliability can fall on long inputs. It also offers an opt-in for sharing queries to help identify model weaknesses, while saying those queries will not be used for training.
““There is still way too much jaggedness for us to vouch for these for production use cases.””
— TypeSafe, in its Jev 1.14 release post
As an affiliate, we earn on qualifying purchases.
News-Image Results Are Missing
The available source text stops partway through the section on news images, after the heading “News images: Jev was right, my images weren’t.” It does not include the test’s news-image scores or explain what Meyer meant by that heading, so those results cannot be reported here. The account also does not give full results for the five essay images or a complete comparison of Jev 1.13.0 and 1.14.0 across all 30 images.
The reported product results are from a small, single-site test, not a public benchmark or independent evaluation. TypeSafe shared internal intelligence estimates rather than public benchmark scores and, according to Meyer, cautioned that the estimates may favor Jev. The release material described here does not establish how performance varies by image quality, subject, task wording or use in production systems.
As an affiliate, we earn on qualifying purchases.
Preview Rollout and Further Testing
TypeSafe said Jev 1.14 would be available through jev-preview immediately after the October 14 release, with jev-latest expected to point to it around October 21. The company’s stated deprecation notice is 60 days for version 1.13.0 and 30 days for version 1.14.0; it said a formal policy was still to come.
For developers, the immediate next step is to test the opt-in features on representative data and check whether confidence scores are reliable for their particular tasks. Meyer’s report would need its missing news-image results, along with broader testing, to support conclusions beyond the product-image sample. TypeSafe has not given a date for removing the experimental warning or for a later release that it considers suitable for production image workflows.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did TypeSafe add in Jev 1.14?
Jev 1.14 adds opt-in image input and a new box response type that returns a bounding box and confidence for a specified target. The box type is available on Jev 1.14; TypeSafe says image input also works with Jev 1.13.0.
Is Jev’s image feature ready for production?
TypeSafe says it is not ready to recommend the feature for production use. Image processing and bounding boxes require explicit opt-in, and the company warned that the results remain inconsistent.
What did Meyer’s 30-image test find?
For 10 product-roundup images, Meyer reports Jev 1.14 matched each image to its own headline more strongly than to the comparison headlines and named the right product in nine of 10 cases. The available account does not include the full results for the 15 news images or five essays.
What image files can Jev accept?
The release details specify PNG and JPEG still images, between 256 and 1,568 pixels per side. The limit is 16 MiB per image and 32 MiB per request; the source says WebP and HEIC files must be converted first.
When will Jev 1.14 become the latest version?
TypeSafe said Jev 1.14 was available as jev-preview after its October 14, 2026 release and that jev-latest would move to it about a week later, around October 21.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
